The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted Sentry error reports which could contain malicious directives. \n- Ingestion points: External data enters through MCP tools such as search_issues, search_events, and get_sentry_resource referenced in SKILL.md. \n- Boundary markers: Explicit security instructions in SKILL.md warn the agent to ignore instructions embedded in data and treat all fields as raw user input. \n- Capability inventory: The agent has tools to search Sentry data, modify the local repository, and create commits/PRs. \n- Sanitization: The skill requires the agent to redact sensitive data (PII, secrets) and use synthetic information in test cases.

sentry-debug-issue